Track and recover jobs

Open a project’s jobs and select a job to inspect its status, progress, target locales, and source file. Use My Jobs to follow your own work across requests.

​

Due dates and ordering

Native jobs can carry a due date set in the workspace, through visual workflow steps (Assign task, Update job), or with the MCP update_job tool (dueAt as ISO 8601, or null to clear). Jobs mirrored from an external TMS show the provider due date when one exists; change those in the TMS, not in Hyperlocalise. On My Jobs and the jobs kanban, open work is sorted with the soonest due date first; jobs without a due date appear after dated work. Cards and the job detail page show relative due text and highlight overdue items. Pick assignee rules can treat urgent work differently using Due within (hours) when the workflow passes a due date into the step.

​

File translation activity

Native file translation and proofread jobs on uploaded project files show an Activity feed on the job detail page. While a translation job is Queued or Running, the page refreshes the feed and status about every three seconds; polling stops in a terminal state. The feed includes:

• Created — when the job was created and by whom, when known.
• Run cards — one card per workflow attempt, labeled with the Hyperlocalise agent. The badge moves through Preparing (context, glossary, translation memory, reading the file), Translating, then Finished, Failed, or Stopped if you cancel or mark the job failed.
• Per locale — waiting, translating, retrying, done, or failed, with translated string counts (for example 120 / 340) and how many strings were reused from translation memory or existing translations.
• Detailed steps — expand a run card to see stages such as strings extracted, translation memory reuse, per-locale progress updates, locale retries, and completion.
• Your actions — cancel, retry, mark as failed, or mark done appear as separate rows when you use those controls. Proofread tasks show who marked the task done.

Jobs started before activity logging was enabled may show only Created and the current status. If activity fails to load, an inline message appears; the rest of the job page still works. The jobs list and kanban do not show this feed yet; open the job detail page to follow progress.

​

Tasks made of strings

A key job is a task for people on a chosen set of strings rather than a file. Create one from the Content Editor: tick strings, open the selection menu, and choose Add to a task…. On a project’s Jobs page, Create job → From strings in the content editor opens the Content Editor ready for ticking. Give a new task a title, languages, and an optional due date and description, or add the strings to an open task of the project. A task covers each of its strings in each of its languages. Automation can create key jobs with the API, picking strings by id, key name, or tag. A key job made with Add to a task… runs no AI translation. Its page shows how many visible strings it holds, and Open Editor opens the project’s Content Editor filtered to the job’s strings. A string hidden after it joined stays in the job but is left out of its count, its progress, and the queue, unless you choose the Hidden filter. Add to a task… in the Content Editor’s selection menu, with Task type set to AI translation, makes a key job of kind translation instead, as Translation does for files. It runs right away: it fills missing translations of its strings in the chosen languages, and Needs review ones that only repeat the source (or, if asked, replaces all not approved yet), saves them as Needs review, and ends Succeeded, listing any languages where some strings failed, or Failed if none could be translated. A failed run can be retried; the retry skips strings already translated. It counts towards the workspace’s open-job limit like other AI jobs, and is charged once, for the job and the AI it used. Reports counts key jobs as it counts file jobs. A key job for people gets its word counts and match bands when it is created, and again when strings are added to it; approvals and edits made inside it count as its review work. An AI translation of chosen strings counts the strings it plans to translate and each one it saves, and its AI use shows under AI usage.

​

Reports on a job

When workspace Reports is enabled, native jobs include a Report tab with that task’s word counts, translation-memory match bands, time logged, and (for administrators and localization managers) cost and pricing. Match analysis runs shortly after the job is created; the tab shows progress until scoring finishes. See Reports.

​

Understand status

Provider-backed jobs can expose different status and action names. Use Open in provider when the connected system owns the job.

​

Retry a job

For a translation job in a queued or failed state, the detail page can offer Retry job. Check the source file, target locales, permissions, usage, and error first. Select Retry job once, then follow the queued attempt. Retrying without fixing a malformed file or disconnected provider is unlikely to help. See Translation failures.

​

Stop an active native job

For a native job that is queued, running, or waiting for review, select Cancel job and confirm when you want to stop it. Cancellation does not undo output or external actions that already completed. Mark as failed is a separate recovery action for active native jobs. It prevents the current workflow from later updating the job and removes it from the active queue. Use it when you have decided to abandon that attempt, rather than treating it as a refresh button.

​

Finish a proofread task

When the work on a proofread task is finished, select Mark done and confirm. The task moves to Succeeded and can’t be reopened. Strings that are not finished stay as they are, and later edits no longer count towards the task. The file language shows as Ready after this, even if some strings are still unfinished. Reviewers, localization managers, and admins can mark any such task done. Anyone else with job access can mark done only a task assigned to them. Post-edit and translate tasks from visual workflows do not have Mark done. Key jobs are finished the same way. Automation can do the same with Complete a task for people.

​

Hand off the result

Use Open Editor when available. Check review permissions if approval is unavailable. Keep the job link when contacting support.